首页> 外文会议>ISCA International Conference on Parallel and Distributed Computing Systems >SPIDER: A MULTI-AGENT ARCHITECTURE FOR INTERNET DISTRIBUTED COMPUTING SYSTEM
【24h】

SPIDER: A MULTI-AGENT ARCHITECTURE FOR INTERNET DISTRIBUTED COMPUTING SYSTEM

机译:蜘蛛:用于互联网分布式计算系统的多代理架构

获取原文
获取外文期刊封面目录资料

摘要

There had been much work done using agents for distributed systems that support E-Commerce and information retrieval systems but practically little or no work has been done for utilizing the massive computing resources available on the Internet to provide a computational platform for solving large complex scientific problems as opposed to running the applications on a supercomputer or a large cluster computing facility. This paper proposes a multi-agent architecture and protocols to support resource sharing and large-scale computation through the Internet. To test the feasibility of the proposed architecture and protocols, we have implemented a prototype using the IBM Aglets. We measured its performance by comparing the running time of a distributed matrix multiplication to a sequential program and the PVM [8]. The results indicate that the multi-agent based architecture can offer better performance for large problem sizes.
机译:使用多大的工作来完成支持电子商务和信息检索系统的分布式系统,而是在互联网上利用可用的大规模计算资源来提供用于解决大型复杂科学问题的计算平台,但实际上很少或没有工作而不是在超级计算机或大型集群计算设施上运行应用程序。本文提出了一种通过互联网支持资源共享和大规模计算的多智能体系结构和协议。要测试所提出的架构和协议的可行性,我们使用IBM Aglet实现了原型。通过将分布式矩阵乘法的运行时间与顺序程序和PVM [8]进行比较来测量其性能。结果表明,基于多种代理的架构可以为大问题尺寸提供更好的性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号